Journal of Instrumentation

OPEN ACCESS

Beam studies of novel THGEM-based potentialsampling elements for Digital Hadron CalorimetryTo cite this article: S Bressler et al 2013 JINST 8 P07017

View the article online for updates and enhancements.

Related contentTHGEM-based detectors for samplingelements in DHCAL: laboratory and beamevaluationL Arazi, H Natal da Luz, D Freytag et al.

-

Recent advances with THGEM detectorsS Bressler, L Arazi, L Moleri et al.

-

First studies with the Resistive-PlateWELL gaseous multiplierA Rubin, L Arazi, S Bressler et al.

-

Recent citationsFirst in-beam studies of a Resistive-PlateWELL gaseous multiplierS. Bressler et al

-

The gain in Thick GEM multipliers and itstime-evolutionM. Alexeev et al

-

p/+ response of single layer THGEMdetector in Ar+3% iC4H10Hong Dao-Jin et al

-

This content was downloaded from IP address 193.137.208.120 on 05/07/2018 at 17:43

2013 JINST 8 P07017

PUBLISHED BY IOP PUBLISHING FOR SISSA MEDIALAB

RECEIVED: May 19, 2013REVISED: June 20, 2013

ACCEPTED: June 26, 2013PUBLISHED: July 29, 2013

Beam studies of novel THGEM-based potentialsampling elements for Digital Hadron Calorimetry

S. Bressler,a,1 L. Arazi,a H. Natal da Luz,b C.D.R. Azevedo,c L. Moleri,a E. Oliveri,d

M. Pitt,a A. Rubin,a J.M.F. dos Santos,b J.F.C.A. Velosoc and A. Breskina

aDept. of Astrophysics and Particle Physics, Weizmann Institute of Science,P.O. Box 26, Rehovot 76100, Israel

bPhysics Department, FCTUC, University of Coimbra,3004-516 Coimbra, Portugal

cI3N, Physics Department, University of Aveiro,3810-193 Aveiro Portugal

dCERN,Meyrin, Switzerland

E-mail: shikma.bressler@weizmann.ac.il

ABSTRACT: Beam studies of thin single- and double-stage THGEM-baseddetectors are presented.Several 10×10cm2 configurations with a total thickness of 5–6 mm (excluding readout electron-ics), with 1×1cm2 pads inductively coupled through a resistive layer to APV-SRS readout elec-tronics, were investigated with muons and pions. Detectionefficiencies in the 98% range wererecorded with an average pad-multiplicity of∼ 1.1. The resistive anode resulted in efficient dis-charge damping, with few-volt potential drops; discharge probabilities were∼ 10−7 for muons and10−6 for pions in the double-stage configuration, at rates of a fewkHz

cm2 . These results, together withthe robustness of THGEM electrodes against spark damage andtheir suitability for economic pro-duction over large areas, make THGEM-based detectors highly competitive compared to the othertechnologies considered for the SiD-DHCAL.

KEYWORDS: Calorimeters; Micropattern gaseous detectors (MSGC, GEM, THGEM, RETHGEM,MHSP, MICROPIC, MICROMEGAS, InGrid, etc); Large detector systems for particle and as-troparticle physics; Gaseous detectors

1Corresponding author.

c© CERN 2013, published under the terms of theCreative Commons Attribution 3.0Licenseby IOP Publishing Ltd and Sissa Medialab srl. Any further distribution of this

work must maintain attribution to the author(s) and the published article’s title, journal citation and DOI.doi:10.1088/1748-0221/8/07/P07017

2013 JINST 8 P07017

Contents

1 Introduction 1

2 Experimental setup and methodology 32.1 The THGEM detectors 32.2 External trigger, tracking and beam parameters 42.3 Readout and data acquisition 52.4 Analysis framework 52.5 Threshold optimization 6

3 Results 73.1 Detection efficiency and pad multiplicity with muons 73.2 Local characteristics (muons) 93.3 Operation with pions 9

4 Discharge analysis 114.1 The effect of micro-discharges 11

5 Summary and discussion 12

1 Introduction

The Thick Gas Electron Multiplier (THGEM) [1] is a simple, robust and economic detector ele-ment; it can be industrially produced over large areas usingstandard Printed Circuit Board (PCB)technologies. Its properties and potential applications are reviewed in [2, 3]; for recent works onTHGEM properties in normal operation conditions see [4–7]. In the last few years, a considerableR&D effort was undertaken to evaluate the applicability of THGEM-based detectors as thin sam-pling elements in Digital Hadronic Calorimeters (DHCAL). The effort focused, in particular, onthe DHCAL of the Silicon Detector (SiD) [8]. However, since THGEM detectors provide propor-tional response, they are also applicable in the semi-DHCALconcept [9]; in the latter, differentthreshold levels applied to the deposited-charge signals should permit reducing hadronic-responsenon-linearities.

SiD is one of the two detector concepts studied for future linear collider, either the InternationalLinear Collider (ILC) [9] or the Compact Linear Collider (CLIC) [10]. In its present design, theSiD DHCAL will require a total active area of a few thousand square meters of a few-millimetresthick sampling elements [8]. The expected average particle rates at the DHCAL are of∼ 1 kHz

cm2 .The ambitious physics program of both the ILC and CLIC requires excellent, 3–4%, jet energy

resolution (defined asσEE , where E is the true jet energy andσE is the resolution of the energy

measurement). In order to achieve this resolution, the SiD DHCAL is designed to employ Particle

– 1 –

2013 JINST 8 P07017

Figure 1. Schematic description of Segmented Resistive Well THGEM (SRWELL) coupled to a readoutpad array.

Flow Algorithms (PFAs) [11], and is thus highly segmented in both the longitudinal and transversedirections. Its baseline design comprises of 40–50 layers of absorber plates (either stainless steelor tungsten) separated by 8 mm gaps, which should incorporate the active sampling elements withtheir 1× 1cm2 pixels and readout electronics [8]. High single-particle detection efficiency andlow average pad-multiplicity (number of pads activated percrossing particle) are essential in thisapplication.

Active elements utilizing 3–5 mm thick Resistive Plate Chambers (RPCs), excluding readoutelectronics, are the baseline technology for the SiD hadronic calorimeter [12, 13]. They were al-ready tested at the level of a 1m3 instrumented block with digital readout electronics and yieldedso far an average pad multiplicity of∼ 1.6 at 94% efficiency with muons [14]. Detection ele-ments based on the MICRO MEsh GAseous Structure (MICROMEGAS), tested with muons at1m2 units, displayed superior properties (with a thickness of∼ 3.1 mm, excluding the MICRO-ROC electronics [15]): 97% efficiency with a 1.1 average multiplicity [16]. Detector prototypeusing 30× 30cm2 double Gas Electron Multipliers (double-GEMs) of 5 mm thickness excludingthe KPiX readout electronics [17], yielded a multiplicity of 1.3 at 95% efficiency with muons [18].

THGEM-based structures were proposed as potential DHCAL sampling elements in 2010 andhave since been thoroughly investigated in laboratory and beam settings. The results of previousbeam studies, performed at the RD51 test-beam facility of CERN-SPS, are summarized in [19].They focused primarily on single- and double-THGEM detector structures of 10× 10cm2, oper-ated in Ne/5%CH4; the avalanche charge was recorded on 1×1cm2 anode readout pads followinga 2 mm induction gap and directly coupled to KPiX readout electronics [17]. These preliminarystudies demonstrated the potential value of such detectorsfor DHCAL, with an average pad mul-tiplicity of ∼ 1.1–1.2 at a muon detection efficiency of∼ 95% for a∼ 6 mm thick single-stagedetector, in which charges were deposited in a 4 mm thick drift gap. These encouraging resultsleft, however, room for optimization, in terms of detector thickness and stability in harsh hadronicenvironment.

A promising step forward for THGEM-based detectors can relyon configurations based on theSegmented Resistive Well THGEM (SRWELL), first suggested in[19]. The SRWELL (figure1) isa composite structure, comprising of a single-faced THGEM electrode (copper-clad on its top sideonly) coupled directly to a resistive anode in a WELL configuration (‘WELL’ being a THGEM withclosed bottom electrode); an avalanche-induced inductivecharge appears on a pad array separated

– 2 –

2013 JINST 8 P07017

from the resistive layer by a thin (100µm) insulating sheet. The absence of the induction gap leadsto a significantly thinner geometry, while the resistive layer serves to effectively quench the energyof occasional discharges. Compared to THGEM configurationswith an induction gap, higher gainsare obtained for lower THGEM voltages, due to the stronger field within the closed holes [19].Charge spreading across the resistive layer may result in delayed inductive signals on neighboringpads and, consequently, in high pad multiplicity. This is prevented by adding a matching grid ofthin copper lines below the resistive layer that allows for rapid clearance of the electrons diffusingover its surface. The SRWELL has a square-hole pattern, with‘blind’ copper strips above the padboundaries, designed to prevent discharges in holes residing directly above the metal grid lines.

We present here recent results of SRWELL-detector evaluation with 150 GeV/c muon andpion beams, conducted at the CERN SPS/H4 RD51 beam-line. Twodetector configurations werestudied: a 5.8–5.9 mm thick single-stage SRWELL detector and a 4.8–6.3 mm thick double-stagemultiplier with a standard THGEM followed by an SRWELL. The detectors were investigatedwith the new Scalable Readout System (SRS) developed withinCERN-RD51 [20]. Results arepresented on detection efficiency, pad multiplicity and discharge probability; future plans are dis-cussed in brief.

2 Experimental setup and methodology

2.1 The THGEM detectors

The THGEM electrodes used in this work were 10×10cm2 in size, manufactured by 0.5 mm diam-eter hole-drilling in either 0.4 or 0.8 mm thick FR4 plates, copper-clad on one or two sides [21]; theholes of the double-sided THGEM electrodes were arranged inan hexagonal lattice with a pitch of1 mm while the holes of the single-sided WELL-THGEM electrode (for the SRWELL configura-tion) were arranged in a square lattice with a pitch of 0.96 mm; 0.1 mm wide rims were chemicallyetched around the holes in both cases. The width of the ‘blind’ copper strips above the grid lines inthe SRWELL electrodes (see figure1) was 0.68 mm. Anodes with surface-resistivity in the range10–20 MΩ/square were used. They were produced by spraying a mixture of graphite powder andepoxy binder on a 100µm thick FR4 sheet, patterned with a square grid of 100µm wide copperlines, defining an array of 10× 10cm2 squares (figure1); the array corresponded to that of the8×8cm2 readout pads, patterned on a 3.2 mm thick FR4 plate. The hole-diameter, electrode thick-ness and rim width were chosen based on previous optimization studies [22]. In view of previousexperience with neon mixtures, offering high attainable gains at relatively low operation poten-tials [6, 23], the detectors were operated in Ne/5%CH4; at normal conditions, 150 GeV muons andpions (referred to in the text as minimally ionizing particles — MIPs) deposit on the average a totalnumber of 60 electrons per cm along their track [24].

The two basic configurations investigated, with different geometrical parameters, are shownin figure2. The single-stage SRWELL detector was investigated with 0.4 and 0.8 mm thick WELLelectrodes; the double-stage detector, with a THGEM preceding the SRWELL, was operated with0.4 mm thick electrodes only. Table1 summarizes the geometrical parameters of both configura-tions and the nominal anode-resistivity values. The drift field was maintained by setting a potentialdifference between the multiplier’s top electrode and an additional copper-plated drift electrode(e.g. a passive THGEM) placed a few mm above.

– 3 –

2013 JINST 8 P07017

Figure 2. Schematic views of the SRWELL detector (a) and the double-stage THGEM+SRWELL (b) con-figurations. Avalanche charges induce signals through a resistive anode, onto pads located behind a 100µmthick insulator. The copper-grid below the resistive film evacuates the electrons diffusing across it surface.

Table 1. Parameters of the single- and double-stage SRWELL detectors (of figure 2) investigated inthis work.

Configuration Thickness Transfer gap Drift gap Total thickness Resistivity[mm] [mm] [mm] [mm] MΩ/square

Single1 0.4 − 5.5 5.9 10

Single2 0.8 − 5 5.8 10

Double1 0.4/0.4 1.5 4 6.3 20

Double2 0.4/0.4 1.5 3 5.3 20

Double3 0.4/0.4 1.5 2.5 4.8 10

The electrodes were polarized with individual HV power-supply CAEN A1833P and A1821Nboards, remotely controlled with a CAEN SY2527 unit. The voltage and current on each channelwere monitored and stored. All inputs were connected through low-pass filters.

2.2 External trigger, tracking and beam parameters

The external trigger system used for event selection is shown in figure3 [25]; it comprised three10× 10cm2 scintillators arranged in coincidence. The tracking system, covering a total area of6× 6cm2, was based on position measurement with three MICROMEGAS detectors, equippedwith an SRS readout system [20] recording signals through APV front-end hybrid electronics [26]similar to the one employed in our test detectors. Two chambers were tested in parallel, containingdifferent detector configurations: one was located betweenthe two downstream trackers and theother between the third tracker and the second scintillator(figure3). The chambers were operatedindependently, but shared the same trigger and tracking system.

The experiments were performed with 150 GeV/c muons and pions. The muon beam wasbroad, covering the entire detector area, with a typical rate of ∼ 10–20 Hz

cm2 . The pion beam hada narrow central peak (∼ 1cm2 width) with wide, low-rate wings. Its rate was varied between

– 4 –

2013 JINST 8 P07017

Figure 3. Schematic description of the test-beam setup, incorporating the two SRWELL detectors, threescintillators and three MICROMEGAS tracking detectors.

∼ 0.5 kHzcm2 and∼ 70 kHz

cm2 to investigate the detector stability and performance under different irradi-ation conditions.

2.3 Readout and data acquisition

A crucial point in the measurements was the synchronizationbetween the data acquired with thedetectors under investigation and that of the tracker. The latter was used to measure the localdetection efficiency and evaluate the level of cross-talk between neighboring pads as function ofthe particle impact location. This was effectively achieved employing the RD51 Scalable ReadoutSystem (SRS) [20].

Like the detectors of the tracker, the 64 readout pads of our detectors were coupled to APV25chips [26]. The APV25, originally designed for the silicon tracking detectors in CMS, has highrate capability and low noise (equivalent of∼ 200 electrons for a few tenth pF typical input capac-itance). Due to the low noise level and high sensitivity of the chip, gas gains of the order of several1000 were sufficient for efficient operation of our detectors.

The data acquisition of both systems was triggered by the same scintillator signals and per-formed with a single SRS front-end card (FEC). For each triggered event, the charge accumulatedby all channels was stored. For each channel, the charge was sampled in 25 bins of 25 ns each.The mmDAQ1 online data acquisition software was used to store the synchronized data on a PCfor further analysis.

2.4 Analysis framework

The track reconstruction algorithm is described in [25]. Tracks were selected by setting a highthreshold on the MICROMEGAS, suppressing noise hits. Thesetracks formed the baseline objects

1Developed by M.Z.D. Byszewski (marcin.byszewski@cern.ch).

– 5 –

2013 JINST 8 P07017

of the analysis; the MIP trajectory through our SRWELL detectors was extrapolated from thecalculated track coordinates. The detector properties, namely pulse-height distribution, efficiencyand pad multiplicity (both global and local) were measured with respect to these trajectories.

Data analysis of the SRWELL detectors comprised the following two steps:

1. Noise suppression: dedicated pedestal runs, without beam, were used to determine the noiselevel of the individual channels in each configuration. The pedestal of each channel wasdefined as the mean value of the charge measured in the 25 ns time intervals. The noiselevel of each channel was defined as the width of this distribution. Only pads with a chargesignal above a threshold relative to the individual channel’s noise were used in the analysisof the physics runs. A change of any of the detector parameters (HV configuration, etc.) wasfollowed by a pedestal run.

2. Clusterization: neighboring pads with detected charge above the thresholdformed a cluster.The cluster position was determined from the average of the pads position, weighted accord-ing to their measured charge. The charge associated with a cluster was taken as the sum ofthe charge of all pads in that cluster.

The detector efficiency was defined as the fraction of tracks where a corresponding clusterwas found with its calculated position not more than 10 mm away from the track trajectory. Thesingle-event multiplicity was defined as the number of pads in the cluster. The average pad mul-tiplicity was defined as the average number of pads in events where a corresponding cluster wasfound not more than 10 mm away from the track trajectory. Boththe detector efficiency and padmultiplicity were measured during stable operation conditions. In particular, time intervals with de-tected high-voltage drops were excluded. Therefore, external parameters such as the power supplyrecovery time following eventual discharges did not bias the measurements. The effect of eventualdischarges on the detector efficiency is discussed below.

The discharge probability was defined here as the number of discharges divided by the numberof hits in the active region of the detector (i.e., in the total area covered by the crossing beam). Thenumber of discharges was extracted directly from the power supply log files by counting the result-ing drops in the monitored voltage. Since the pion beam was narrower than both the acceptance ofthe scintillators and the detectors, the number of pion hitsin the active region of the detector wasestimated as the number of triggers. For muons, where the beam was wider than the acceptance ofboth the scintillators and the detectors, the number of muonhits in the active area of the detectorwas estimated from the number of triggers, by correcting forthe difference between the acceptanceof the scintillators (6×6cm2) and that of the detector (10×10cm2).

2.5 Threshold optimization

Unlike the tracker, where a maximal threshold was set duringthe analysis to ensure that only good-quality tracks are used, the threshold for the SRWELL detectors was set to optimize the detectorperformance. Figure4 shows an example of global detector efficiency recorded withmuons, versusthe pad-multiplicity; the data were recorded with a single-stage 0.8 mm thick SRWELL detector(figure 2a; Single2in table1) for different thresholds set during the analysis. The detector bias

– 6 –

2013 JINST 8 P07017

Figure 4. Global detection efficiency as a function of the pad-multiplicity for different charge thresholds setin the analysis, as measured with a 0.8 thick SRWELL (figure2a) with muons. Ne/5%CH4;∆VDrift = 250 V.

voltage (∆VSRWELL) was scanned while the drift field was kept constant (∆VDrift = 250 V corre-sponding to a drift field of approximately 0.5 kV/cm); as explained in section 2.4, the thresholdswere set relative to the noise level of the individual channels. As can be seen, relative thresholds of0.7 or 0.9 were optimal in this case; they resulted in high detection efficiency and low pad multi-plicity. A threshold of 0.7 was used throughout the analysissince it was found optimal also for theother SRWELL detector configurations investigated here.

3 Results

3.1 Detection efficiency and pad multiplicity with muons

The detector efficiency as a function of the applied voltage is shown in figure5 for theTHGEM+SRWELL in aDouble1configuration (table1). In figure 5a the same potentials wereapplied across the THGEM and the SRWELL (∆VTHGEM = ∆VSRWELL). This potential was var-ied keeping a constant transfer potential:∆VTransfer= 200 V (corresponding to a transfer field ofapproximately 1.3 kV/cm); in figure5b the transfer field was varied, keeping a constant potentialdifferences across both multipliers:∆VSRWELL = ∆VTHGEM = 560 V. Conditions were found forreaching close-to-unity efficiencies in a stable operationmode.

The results of the measured detection efficiency with muons as a function of the pad multiplic-ity for the five configurations investigated (table1) are summarized in figure6. All measurementswere done under stable operation conditions. For both theSingle1andSingle2configurations, themeasurements were done by increasing the SRWELL voltage; for Double1andDouble3config-urations the transfer voltage was kept constant (∆VTransfer= 300 V and∆VTransfer= 50 V respec-

– 7 –

2013 JINST 8 P07017

Figure 5. The detection efficiency with muons of the THGEM+SRWELL in aDouble1configuration (fig-ure2b; table1) as a function of applied potential. (a) Efficiency as a function of the voltage across both mul-tipliers (∆VSRWELL = ∆VTHGEM) at constnt∆VTransfer= 200 V; (b) efficiency as a function of the∆VTransfer

at constant∆VSRWELL = ∆VTHGEM = 560 V.

Figure 6. Detection efficiency with muons as a function of the pad multiplicity measured with the single-SRWELL and the double THGEM+SRWELL configurations listed intable1. The total detector thickness,excluding readout electronics, is provided for each case inthe figure. The optimal voltages and resultingefficiency, multiplicity and gain values are given in table2.

tively, or a transfer fields of approximately 2 kV/cm and 0.3 kV/cm) while the bias voltages of theTHGEM and SRWELL were increased symmetrically (keeping∆VSRWELL = ∆VTHGEM); for Dou-ble2 the transfer voltage was kept constant (∆VTransfer= 225 V, or approximately transfer field of

– 8 –

2013 JINST 8 P07017

Table 2. The applied potentials and resulting effective gain at theoptimal detection conditions (lowest padmultiplicity at detection efficiency plateau) of all of the tested configurations.

Configuration ∆VSRWELL ∆VTransfer ∆VTHGEM Efficiency Multiplicity Effective[V] [V] [V] [mm] gain

Single1 610 − − 0.97 1.2 1200

Single2 780 − − 0.98 1.1 2000

Double1 530 300 530 0.98 1.2 6500

Double2 540 225 540 0.95 1.2 8200

Double3 540 50 540 0.98 1.3 4000

1.5 kV/cm) while the bias voltages of the THGEM and SRWELL were scanned asymmetrically.High detection efficiencies (> 95%) and low pad multiplicity values (< 1.3) were recorded withall configurations. Note that over 98% efficiency at multiplicity values below 1.2 were measuredwith theSingle2andDouble1detector configurations. The applied voltages and resulting gains atoptimal detection conditions (lowest pad multiplicity at the efficiency plateau) are summarized forall configurations in table2.

3.2 Local characteristics (muons)

The accurate tracking allowed for studying the local characteristics of the detector. For each con-figuration, an optimal working point (lowest multiplicity at the efficiency plateau) was selected. Atthese conditions, the local efficiency and multiplicity were measured as a function of the particleimpact distance from the edge of the pad. The local efficiencyand multiplicity plots of theSingle2andDouble1detector configurations are shown in figure7. As expected, close to the edge of thepad the charge is shared between two neighboring pads. Similar behavior was measured with theother three configurations.

As discussed in our previous work [19] the charge sharing also depends on the detector con-figuration, the anode type and its resistivity; in the SRWELL, the metal strips below the resistivelayer prevent charge propagation to neighboring pads and the charge sharing is due mostly to theevent induced avalanche size. By definition, the charge sharing between neighboring pads results inhigher average pad multiplicity. This suggests that the padmultiplicity provides indeed additionalinformation concerning the track position, which could be exploited. Since in such events, lesscharge is induced on each pad, the signal-to-noise separation is worse, resulting in slightly lowerdetection efficiency.

3.3 Operation with pions

Experiments were carried out with 150 GeV/c pions, to investigate the detectors’ behavior underhigher event rate and in the presence of potential higher-ionization hadron-induced background.Examples of cluster charge distributions (defined in section 2.4) measured with both muons andpions in theSingle2andDouble1configurations are shown in figure8 and figure9 respectively.For each configuration the measurements were carried out with the same high-voltage configura-tion for both beam types;∆VSRWELL = 800 V for Single2. ∆VSRWELL = ∆VTHGEM = 560 V and

– 9 –

2013 JINST 8 P07017

Figure 7. Local efficiency and multiplicity data measured with the Single2 SRWELL and the Double1THGEM+SRWELL, using the tracking system. (a) Detection efficiency as a function of distance from theedge of the pad. (b) Pad multiplicity as a function of distance from the edge of the pad. The operationpotentials are depicted in table2.

Figure 8. Cluster charge distributions measured with theSingle2configuration (table1) at an operationvoltage∆VSRWELL = 800 V (a) muons, 10–20Hz

cm2 ; (b) pions, 0.5kHzcm2 . The measured data (blue histograms)

are fitted to Landau distributions.

∆VTransfer= 50 V for Double1. The measured data (blue histograms) are fitted to Landau dis-tributions (red curves). As can be seen, the measured cluster charge distribution of the double-stage configuration (figure2b) remained practically unchanged in the transition from muons topions; however, the operation of the single-stage configuration (figure2a) with pions resulted inan average-gain drop of over 50%, as indicated by the drop of the Landau Most Probable Value(MPV) from 6.6 fC to 3.1 fC.

– 10 –

2013 JINST 8 P07017

Figure 9. Cluster charge distributions measured with theDouble1configuration (table1) at operation volt-ages∆VSRWELL = 560 V, ∆VTHGEM = 560 V and∆VTransfer = 50 V. (a) Muons, 10–20Hz

cm2 ; (b) pions,

0.5 kHzcm2 . The measured data (blue histograms) are fitted to Landau distributions.

The substantial drop of the MPV in the single-SRWELL irradiated with pions (figure8b) is yetunclear, requiring further investigations. The possibility of rate-dependence of the gain is ruled out,since this gain drop is not reproduced in the lab when irradiating the detector with an X-ray sourceat the same rates. Note that some high-voltage fluctuations that appeared at the much higher ratesin the single-stage configuration operated with pions couldhave also caused this effect. Carefulstudies are in course.

4 Discharge analysis

In previous laboratory studies, occasional-discharge rates and amplitudes measured with a resistive(but not segmented) WELL were considerably lower compared to that of THGEM and WELLconfigurations coupled to metal pads [19]. In the present beam study, two beam-related types ofdischarges were observed. Large voltage drops of 50–150 V ateither one or more of the detectorelectrodes, characterized the first type; a few volts drop characterized the second type, referred inthe text as micro-discharges.

The probability of large discharges was low,< 10−6, for all of the investigated configurations.They depended on the detector configuration, on the applied voltage and on the particle flux. Inparticular, no large discharges of the SRWELL electrode were observed with the double-stageconfigurations. Large discharges were followed by long recovery times of the power supply andpossibly of the detector. In some cases, the SRS readout electronics had to be reconfigured orpower-cycled.

4.1 The effect of micro-discharges

Typical high-voltage variations in time, measured on the different electrodes of a double-stageconfiguration (Double3) operated in a pion beam, are shown in figure10. In order to study the

– 11 –

2013 JINST 8 P07017

Figure 10. TheDouble3THGEM+SRWELL detector configuration. High-voltage variation in time overlaidon the spill structure, as measured in a pion beam at a rate of 3kHz

cm2 . No high voltage variations were measuredon the mesh (magenta curve). The small variations (shown as dots) measured on the THGEM top electrode,THGEM bottom electrode and WELL electrode are shown in black, green and red curves respectively.

correlation between the occasional voltage drops and the beam flux, the variations are overlaid onthe number of tracks measured every ten seconds (the beam consisted of 10 seconds spills every50 seconds). As can be seen, the THGEM top and SRWELL electrodes are those with the highestactivity. The THGEM top displays 50–100 V discharges, whilethe SRWELL shows only micro-discharges (∼ 3 V drops). Discharges are mainly, but not exclusively, correlated with the beam.Discharges on the THGEM top and micro-discharges on the SRWELL are partially correlated.

The performance of the double-stage configuration (Double3) was measured during pion-beamspills, with and without the appearance of micro-discharges. Similar (within statistical fluctuation)efficiencies were measured in both cases. The very similar cluster-charge distributions recordedin both conditions are shown in figure11. We conclude that the effect of micro-discharges on theperformance of the double-stage detector is negligible. Similar analysis could not be made with thesingle-stage detectors since it was difficult to find spills with and without high-voltage drops at thesame high-voltage configuration.

5 Summary and discussion

This work focused on the test-beam results with our thinnest(< 6.3 mm) THGEM/SRWELL de-tectors, as possible candidates for Digital Hadron Calorimetery applications. A total of five config-urations were investigated; two single-stage SRWELL and three double-stage THGEM+SRWELLconfigurations. All the detectors were operated in stable conditions at typical effective gains in therange 1000–8000; signals were processed with frontend APV chips and SRS electronics. Detailedresults of systematic studies of processes in the differentdetector configurations discussed here willbe the subject of forthcoming publications.

– 12 –

2013 JINST 8 P07017

Figure 11. Cluster charge distributions recorded in theDouble3THGEM+SRWELL detector (of figure10)with pions at a rate of 3kHz

cm2 . The distributions were measured during spills with (red) and without (blue)micro-discharges in the same operation conditions:∆VSRWELL = 550 V,∆VTHGEM = 550 V and∆VTransfer=

50 V resulting in an effective gain of∼ 2500.

High detection efficiencies (over 95%) and low average pad-multiplicity values (less than 1.2)were recorded with all the configurations investigated here. In particular, efficiencies above 98%with an average pad-multiplicity below 1.2 were recorded with both theSingle2andDouble1con-figurations, with a total thickness of 5.8 and 6.3 mm respectively. In the transition from low-ratemuon-beam to high-rate pion beam a significant, yet unclear,gain drop was observed with thesingle-stage configuration, at the same operation voltage;the effect is being investigated. No gaindrop was observed with the double-stage configurations.

Two types of occasional discharges were observed; large discharges, characterized by a voltagedrops of 50–150 V, were observed with the single-stage configurations at low appearance rates(< 10−6), and very rarely on the top THGEM electrode of the double-stage configurations. Micro-discharges, characterized by a voltage drop smaller than 10V, were recorded in all configurations;these did not affect the cluster-charge distributions and the detection efficiency of the double-stageconfigurations. Further studies are required in order to understand the effect of micro-dischargeson the single-stage configuration.

Compared to the technologies already explored as potentialsampling-elements for future Dig-ital Hadron Calorimeters, the performance of the 10× 10cm2 THGEM-based detectors reportedin this work is better than that of 1× 1m2 RPCs and 30× 30cm2 GEM detectors and similar tothat of 1×1m2 MICROMEGAS. The detector thickness can be further reduced by optimizing thedifferent gaps. The thinner single-stage configuration hasadditional advantages in term of cost andcomplexity of the detector.

The proportional response of our THGEM-based sampling elements makes them attractivealso for the semi-DHCAL concept. Note that preliminary investigations of a single-THGEM and

– 13 –

2013 JINST 8 P07017

SRWELL yielded satisfactory results, in terms of efficiencyand pad multiplicity,2 employing theMICROROC readout electronics; the latter (developed for DHCAL applications [15]) can be oper-ated also in dual-threshold semi-digital mode.

The challenging task of developing a large-scale thin single-stage detector also requires re-solving the observed gain-drop effect. It will be the focus of our future work. The assembly oflarger 30×30cm2 SRWELL-detector prototypes is in course.

Acknowledgments

We are indebted to Hans Muller, Sorin Martoiu, Marcin Byszewski and Konstantinos Karakostasfor their assistance with the SRS electronics and associated software. We thank Victor Revivo andAmnon Cohen of Print Electronics Israel for producing the detector electrodes. We thank Rui deOliveira of CERN for helpful discussions and assistance with the detector electrodes, for producingthe readout pad array and for his on-site support during the beam test. We thank Andy White ofUTA for his interest in this work and his support and Leszek Ropelewski and his team at CERN’sGas Detector Development group, in particular George Glonti, for their kind assistance duringthe tests. This work was supported in part by the Israel-U.S.A. Binational Science Foundation(Grant 2008246), by the Benozyio Foundation and by the FCT Projects PTDC/FIS/113005/2009and CERN/FP/116394/2010. The research was done within the CERN RD51 collaboration. H. Na-tal da Luz is supported by FCT grant SFRH/BPD/66737/2009. C.D.R. Azevedo is supported bythe SFRH/BPD/79163/2011 grant. A. Breskin is the W.P. Reuther Professor of Research in thePeaceful use of Atomic Energy.

References

[1] R. Chechik, A. Breskin, C. Shalem and D. Mormann,Thick GEM-like hole multipliers: propertiesand possible applications, Nucl. Instrum. Meth.A 535 (2004) 303[physics/0404119].

[2] A. Breskin et al.,A concise review on THGEM detectors, Nucl. Instrum. Meth.A 598 (2009) 107[arXiv:0807.2026].

[3] A. Breskin et al.,CsI-THGEM gaseous photomultipliers for RICH and noble-liquid detectors,Nucl. Instrum. Meth.A 639 (2011) 117[arXiv:1009.5883].

[4] V. Peskov, M. Cortesi, R. Chechik and A. Breskin,Further evaluation of a THGEM UV-photondetector for RICH — comparison with MWPC, 2010JINST5 P11004.

[5] M. Alexeev et al.,Development of THGEM-based photon detectors for COMPASS RICH-1,Phys. Procedia37 (2012) 781.

[6] C.D.R. Azevedo et al.,Towards THGEM UV-photon detectors for RICH: on single-photon detectionefficiency in Ne/CH4 and Ne/CF4, 2010JINST5 P01002.

[7] M. Alexeev et al.,Ion backflow in thick GEM-based detectors of single photons,2013JINST8 P01021.

[8] P.N. Burrows, L. Linssen, M. Oreglia, M. Stanitzki and A.White,SiD detailed baseline design,http://www.linearcollider.org/ILC/physics-detectors/Detectors.

2A collaborative ongoing effort with LAPP Annecy.

– 14 –

2013 JINST 8 P07017

[9] J. Brau, Y. Okada and N. Walker,The International Linear Collider technical design report,http://www.linearcollider.org/ILC/Publications/Technical-Design-Report(2013).

[10] L. Linssen, A. Miyamoto, M. Stanitzki and H. Weerts,Physics and detectors at CLIC: CLICconceptual design report, arXiv:1202.5940.

[11] L. Xia, Development of a particle flow algorithm at Argonne, in Proceedings of theXII InternationalConference on Calorimetry in High Energy Physics, Chicago U.S.A., 5–9 Jun 2006[AIP Conf. Proc.867(2006) 531].

[12] G. Drake, J. Repond, D.G. Underwood and L. Xia,Resistive Plate Chambers for hadron calorimetry:tests with analog readout, Nucl. Instrum. Meth.A 578 (2007) 88.

[13] B. Bilki et al., Measurement of the rate capability of Resistive Plate Chambers,2009JINST4 P06003.

[14] B. Bilki et al., Calibration of a digital hadron calorimeter with muons, 2008JINST3 P05001.

[15] C. Adloff et al.,MICROROC: MICRO-mesh gaseous structure Read-Out Chip,2012JINST7 C01029.

[16] C. Adloff et al.,MICROMEGAS chambers for hadronic calorimetry at a future linear collider,2009JINST4 P11023.

[17] D. Freytag et al.,KPiX, an array of self triggered charge sensitive cells generating digital time andamplitude information, IEEE Nucl. Sci. Symp. Conf. Rec.(2008) 3447.

[18] J. Yu et al.,Application of large scale GEM for digital hadron calorimetry,Phys. Procedia37 (2012) 393[arXiv:1110.4326].

[19] L. Arazi et al.,THGEM-based detectors for sampling elements in DHCAL: laboratory and beamevaluation, 2012JINST7 C05011.

[20] S. Martoiu, H. Muller, A. Tarazona and J. Toledo,Development of the scalable readout system formicro-pattern gas detectors and other applications, 2013JINST8 C03015.

[21] Print Electronics ltd.,http://www.print-e.co.il/home-page.asp.

[22] C. Shalem, R. Chechik, A. Breskin and K. Michaeli,Advances in thick GEM-like gaseous electronmultipliers. Part I: Atmospheric pressure operation, Nucl. Instrum. Meth.A 558 (2006) 475[physics/0601115].

[23] M. Cortesi et al.,THGEM operation in Ne and Ne/CH4, 2009JINST4 P08001.

[24] PARTICLE DATA GROUPcollaboration, J. Beringer et al.,Review of particle physics,Phys. Rev.D 86 (2012) 010001.

[25] K. Karakostas et al.,Micromegas telescope for the RD51 test beams, in Proceedings of theXXVIIIWorkshop on Recent Advances in Particles Physics and Cosmology, Thessaloniki Greece,25–28 Mar 2010,http://hep2010.physics.auth.gr/.1

[26] M. French et al.,Design and results from the APV25, a deep sub-micron CMOS front-end chip for theCMS tracker, Nucl. Instrum. Meth.A 466 (2001) 359.

– 15 –